A novel LBP fuzzy feature extraction method for face recognition

This paper presents theoretically simple, yet computationally efficient approach for face recognition. In this approach the face image is divided into several sub-regions from which the information derived using the Local Binary Pattern (LBP) over a window and the information at the central pixel, which is a product of information source and fuzzy membership value. The LBP features possess the texture discriminative property with low computational cost. By taking the information from both LBP and the central pixel, the shortcoming of LBP is removed. The proposed approach fares well over PCA and LDA when implemented using SVM and KNN as the classifiers on ORL database and CSIST database.